After 10 Years, NIST Physicist Stephan Schlamminger Opens His Sealed Envelope and Finds That His New Big G Measurement Disagrees With the French Result by 2.3 Parts in 10,000, Deepening a 225-Year-Old Mystery

GAITHERSBURG, Md. — After 10 years of painstaking measurements, the use of a tungsten torsion balance the length of a small car, and an unusual scientific ritual of keeping his own preliminary results sealed inside an envelope so he could not subconsciously bias them, physicist Stephan Schlamminger of the National Institute of Standards and Technology has produced what is now the most precise American measurement of the universal gravitational constant — and the result, far from settling a centuries-old mystery, has only deepened it.

For more than 225 years, since Henry Cavendish first attempted in 1798 to weigh the Earth using a torsion balance, physicists have tried to pin down the value of "big G," the constant that relates the mass of two objects to the gravitational pull between them. Every other constant in physics — the speed of light, Planck's constant, the elementary charge — is known to roughly 10 parts per billion or better. Big G is known to about one part in 10,000, and the most precise modern measurements disagree with one another by more than the experimenters' stated uncertainties. "The discrepancy is real, it is persistent, and it has gotten worse, not better, as we have built more sensitive instruments," Schlamminger said in a CNN interview published this month.

Schlamminger's NIST team recreated a landmark 2001 experiment by the Bureau International des Poids et Mesures in France, using two large stainless-steel weights suspended on a thin tungsten fiber inside a vacuum chamber. To prevent unconscious bias, his collaborators encrypted a critical calibration constant in a sealed envelope at the start of the experiment in 2016 and refused to open it until the analysis was complete. When Schlamminger finally broke the seal in late April, his team's value came in 0.0235 percent lower than the original French result — a difference more than twice the combined experimental uncertainty.

"You hope that when you finally open that envelope you'll see your number agree with everybody else's, and the answer to a 225-year-old question would be solved," Schlamminger told Scientific American in a May feature. "That is not what happened." The new NIST value, formally G = 6.67430 × 10⁻¹¹ m³ kg⁻¹ s⁻² with a relative uncertainty of 22 parts per million, was published Tuesday in the journal Physical Review Letters and announced in a paper at the spring meeting of the American Physical Society in Anaheim.

The persistent disagreement among measurements is one of the most stubborn unresolved problems in fundamental metrology. Some theorists have speculated that the spread might hint at new physics — perhaps a slow time variation in the constant, or a fifth force operating at laboratory scales — though most experimentalists, including Schlamminger, believe the answer is more likely an unknown systematic error common to one or more of the existing apparatuses. NIST is now collaborating with teams at the University of Washington and the International Bureau of Weights and Measures in Sèvres, France, on a coordinated set of cross-comparisons using both torsion balances and atom-interferometry techniques scheduled to begin in 2027. Until then, big G will remain the most precisely studied and least precisely known number in physics.